Sqs

Alloy Theoretic Automated Toolkit (ATAT)
1

I am trying to generate SQS of a structure with Pearson Symbol oI158. It has 20 Wyckoff positions, but some sites are partially occupied as revealed X-ray diffraction. This is why I want to generate SQS supercell so that total energy could be calculated from first-principles.

So as a first attempt I generated a conventional unit cell, and ran
cellcvrt -s < input > output
Unfortunately, the output give me only atom positions with the fist Wyckoff site in the input file!

Any idea what, if any, went wrong!
Thanks,
Gautam

2

A couple of issues in your input files:

  1. even if the atoms occupying two consecutive sites in the file are the same you still need to repeat them on each line.
  2. If you want the code to be able to regroup similar sites, they should have the same species
    For instance something like:
...
0.614800 0.000000 0.000000 Mg1at4e
0.385200 0.000000 0.000000
0.114800 0.500000 0.500000
0.885200 0.500000 0.500000
...

should be

...
0.614800 0.000000 0.000000 Mg
0.385200 0.000000 0.000000 Mg
0.114800 0.500000 0.500000 Mg
0.885200 0.500000 0.500000 Mg
...
  1. Another issue is that your input file contains some overlapping sites, which you can remove with cellcvrt -rr < str.txt > str.out
    (Assuming str.in is your structure corrected as above - I am attaching the corrected file.)
    Then you can type:
    cellcvrt -s < str.out
    and it does find a smaller cell (by factor 2 only).
    In case you wanted to Zn and Mg sites to be considered equivalent then do
    sed ‘s/Mg/Zn/g’ < str.out > str2.out
    cellcvrt -s < str2.out
    This did yield any further reduction in this case.
3

Hi Axel,
I am trying to generate ternary bcc-SQS, can you please confirm if my lat.in and rndstr.in are OK. The first line represents a, b, c, alpha, beta, gamma, but it is followed by four lines? … something I did not understand.

--------------lat.in------------
3.0 3.0 3.0 90 90 90
0.5 0.5 0.5
0.5 -0.5 0.5
0.5 0.5 -0.5
0 0 0 A,B,C

--------------rndstr.in----------
3.0 3.0 3.0 90 90 90
0.5 0.5 0.5
0.5 -0.5 0.5
0.5 0.5 -0.5
0 0 0 A=0.25,B=0.50,C=0.25

Thanks,
Gautam

4

I’m very sorry for the very delayed answer! (It is proposal writing season and I have a high teaching load this semester!)

Your rndstr.in file looks correct.
NOTE: With the new code mcsqs you don’t need lat.in only the rndstr.in input file.
The corrdump code can read the rndstr.in file if you indicate the options: -ro -l=rndstr.in

5

Hi Axel,
I am trying to generate SQS of ternary alloy. I have created rndstr.in, and then executed
corrdump -l=rndstr.in -ro -noe -nop -clus -2=5.0 -3=3.0

The bestcorr.out gives me

2 2.598076 0.062500 0.062500 -0.000000
2 2.598076 0.054127 0.054127 -0.000000
2 2.598076 0.046875 0.046875 -0.000000
2 3.000000 0.062500 0.062500 -0.000000
2 3.000000 0.040595 0.054127 -0.013532
2 3.000000 0.046875 0.046875 0.000000
2 4.242641 0.062500 0.062500 -0.000000
2 4.242641 0.054127 0.054127 -0.000000
2 4.242641 0.031250 0.046875 -0.015625
2 4.974937 0.062500 0.062500 -0.000000
2 4.974937 0.060892 0.054127 0.006766
2 4.974937 0.046875 0.046875 -0.000000
3 3.000000 -0.024414 -0.015625 -0.008789
3 3.000000 -0.015223 -0.013532 -0.001691
3 3.000000 0.001691 -0.013532 0.015223
3 3.000000 -0.008789 -0.011719 0.002930
3 3.000000 -0.014648 -0.011719 -0.002930
3 3.000000 -0.015223 -0.010149 -0.005074

Looking in this list, is it fair to say -2=5.0 -3=3.0 options yield 3 pairs and 1 triplet (distance wise, 1st column in above table), or 12 pairs and 6 triplets? What is the correct statement here?

Thanks,
Gautam

6

The only way to be sure is to look into the clusters.out file. Those clusters that have the same coordinates (1st 3 numbers) but different "functions" (column 4 and 5) should count as one cluster.
Lenght is not sufficient to decide on equivalence (clearly for 3-point clusters but also possibly for pairs if the structure is complex).